diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7186 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998075684177941,
+  "global_step": 597500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9983266818938615e-05,
+      "loss": 1.0931,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.996653363787723e-05,
+      "loss": 1.0437,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9949800456815846e-05,
+      "loss": 0.9591,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.993306727575446e-05,
+      "loss": 0.9321,
+      "step": 2000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9916334094693073e-05,
+      "loss": 0.9527,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9899600913631687e-05,
+      "loss": 0.9186,
+      "step": 3000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98828677325703e-05,
+      "loss": 0.9105,
+      "step": 3500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9866134551508914e-05,
+      "loss": 0.9094,
+      "step": 4000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.984940137044753e-05,
+      "loss": 0.9226,
+      "step": 4500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9832668189386145e-05,
+      "loss": 0.9058,
+      "step": 5000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.981593500832476e-05,
+      "loss": 0.9121,
+      "step": 5500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9799201827263376e-05,
+      "loss": 0.8985,
+      "step": 6000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.978246864620199e-05,
+      "loss": 0.8825,
+      "step": 6500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9765735465140603e-05,
+      "loss": 0.8887,
+      "step": 7000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9749002284079217e-05,
+      "loss": 0.8718,
+      "step": 7500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.973226910301783e-05,
+      "loss": 0.8863,
+      "step": 8000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9715535921956447e-05,
+      "loss": 0.8953,
+      "step": 8500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.969880274089506e-05,
+      "loss": 0.876,
+      "step": 9000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9682069559833675e-05,
+      "loss": 0.8526,
+      "step": 9500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9665336378772288e-05,
+      "loss": 0.8793,
+      "step": 10000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9648603197710902e-05,
+      "loss": 0.8571,
+      "step": 10500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9631870016649516e-05,
+      "loss": 0.8752,
+      "step": 11000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9615136835588133e-05,
+      "loss": 0.856,
+      "step": 11500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9598403654526746e-05,
+      "loss": 0.861,
+      "step": 12000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.958167047346536e-05,
+      "loss": 0.8542,
+      "step": 12500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9564937292403974e-05,
+      "loss": 0.8354,
+      "step": 13000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9548204111342587e-05,
+      "loss": 0.8476,
+      "step": 13500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.95314709302812e-05,
+      "loss": 0.8695,
+      "step": 14000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9514737749219818e-05,
+      "loss": 0.8872,
+      "step": 14500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.949800456815843e-05,
+      "loss": 0.8373,
+      "step": 15000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9481271387097045e-05,
+      "loss": 0.849,
+      "step": 15500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.946453820603566e-05,
+      "loss": 0.8236,
+      "step": 16000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9447805024974273e-05,
+      "loss": 0.845,
+      "step": 16500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9431071843912886e-05,
+      "loss": 0.8399,
+      "step": 17000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9414338662851503e-05,
+      "loss": 0.8132,
+      "step": 17500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9397605481790117e-05,
+      "loss": 0.8424,
+      "step": 18000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9380872300728734e-05,
+      "loss": 0.8348,
+      "step": 18500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9364139119667348e-05,
+      "loss": 0.8318,
+      "step": 19000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.934740593860596e-05,
+      "loss": 0.8304,
+      "step": 19500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9330672757544575e-05,
+      "loss": 0.8388,
+      "step": 20000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.931393957648319e-05,
+      "loss": 0.837,
+      "step": 20500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9297206395421802e-05,
+      "loss": 0.8213,
+      "step": 21000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.928047321436042e-05,
+      "loss": 0.8347,
+      "step": 21500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9263740033299033e-05,
+      "loss": 0.8231,
+      "step": 22000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9247006852237647e-05,
+      "loss": 0.8212,
+      "step": 22500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.923027367117626e-05,
+      "loss": 0.8117,
+      "step": 23000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9213540490114874e-05,
+      "loss": 0.8134,
+      "step": 23500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9196807309053487e-05,
+      "loss": 0.823,
+      "step": 24000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9180074127992105e-05,
+      "loss": 0.8208,
+      "step": 24500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9163340946930718e-05,
+      "loss": 0.83,
+      "step": 25000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9146607765869332e-05,
+      "loss": 0.8225,
+      "step": 25500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9129874584807945e-05,
+      "loss": 0.8116,
+      "step": 26000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.911314140374656e-05,
+      "loss": 0.8098,
+      "step": 26500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9096408222685173e-05,
+      "loss": 0.8269,
+      "step": 27000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.907967504162379e-05,
+      "loss": 0.8196,
+      "step": 27500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9062941860562403e-05,
+      "loss": 0.8185,
+      "step": 28000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9046208679501017e-05,
+      "loss": 0.8318,
+      "step": 28500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9029475498439634e-05,
+      "loss": 0.8227,
+      "step": 29000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9012742317378248e-05,
+      "loss": 0.8182,
+      "step": 29500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.899600913631686e-05,
+      "loss": 0.8062,
+      "step": 30000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8979275955255475e-05,
+      "loss": 0.8161,
+      "step": 30500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.896254277419409e-05,
+      "loss": 0.7953,
+      "step": 31000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8945809593132706e-05,
+      "loss": 0.8159,
+      "step": 31500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.892907641207132e-05,
+      "loss": 0.807,
+      "step": 32000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8912343231009933e-05,
+      "loss": 0.8216,
+      "step": 32500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8895610049948547e-05,
+      "loss": 0.8106,
+      "step": 33000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.887887686888716e-05,
+      "loss": 0.7772,
+      "step": 33500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8862143687825774e-05,
+      "loss": 0.813,
+      "step": 34000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.884541050676439e-05,
+      "loss": 0.8245,
+      "step": 34500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8828677325703005e-05,
+      "loss": 0.7687,
+      "step": 35000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.881194414464162e-05,
+      "loss": 0.7935,
+      "step": 35500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8795210963580232e-05,
+      "loss": 0.7943,
+      "step": 36000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8778477782518846e-05,
+      "loss": 0.7958,
+      "step": 36500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8761744601457463e-05,
+      "loss": 0.7948,
+      "step": 37000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8745011420396076e-05,
+      "loss": 0.7991,
+      "step": 37500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.872827823933469e-05,
+      "loss": 0.8116,
+      "step": 38000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8711545058273304e-05,
+      "loss": 0.78,
+      "step": 38500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8694811877211917e-05,
+      "loss": 0.7934,
+      "step": 39000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.867807869615053e-05,
+      "loss": 0.8229,
+      "step": 39500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8661345515089148e-05,
+      "loss": 0.8017,
+      "step": 40000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8644612334027762e-05,
+      "loss": 0.7874,
+      "step": 40500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.862787915296638e-05,
+      "loss": 0.7789,
+      "step": 41000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8611145971904992e-05,
+      "loss": 0.8207,
+      "step": 41500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8594412790843606e-05,
+      "loss": 0.8013,
+      "step": 42000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.857767960978222e-05,
+      "loss": 0.7669,
+      "step": 42500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8560946428720833e-05,
+      "loss": 0.795,
+      "step": 43000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8544213247659447e-05,
+      "loss": 0.7827,
+      "step": 43500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8527480066598064e-05,
+      "loss": 0.7945,
+      "step": 44000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8510746885536678e-05,
+      "loss": 0.7918,
+      "step": 44500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.849401370447529e-05,
+      "loss": 0.7635,
+      "step": 45000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8477280523413905e-05,
+      "loss": 0.7738,
+      "step": 45500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.846054734235252e-05,
+      "loss": 0.7885,
+      "step": 46000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8443814161291132e-05,
+      "loss": 0.8243,
+      "step": 46500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.842708098022975e-05,
+      "loss": 0.7857,
+      "step": 47000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8410347799168363e-05,
+      "loss": 0.7914,
+      "step": 47500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8393614618106977e-05,
+      "loss": 0.7951,
+      "step": 48000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.837688143704559e-05,
+      "loss": 0.7843,
+      "step": 48500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8360148255984204e-05,
+      "loss": 0.8035,
+      "step": 49000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8343415074922818e-05,
+      "loss": 0.799,
+      "step": 49500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8326681893861435e-05,
+      "loss": 0.8047,
+      "step": 50000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.830994871280005e-05,
+      "loss": 0.7825,
+      "step": 50500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8293215531738662e-05,
+      "loss": 0.8013,
+      "step": 51000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8276482350677276e-05,
+      "loss": 0.7901,
+      "step": 51500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8259749169615893e-05,
+      "loss": 0.771,
+      "step": 52000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8243015988554506e-05,
+      "loss": 0.7974,
+      "step": 52500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.822628280749312e-05,
+      "loss": 0.7891,
+      "step": 53000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8209549626431734e-05,
+      "loss": 0.7718,
+      "step": 53500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.819281644537035e-05,
+      "loss": 0.8089,
+      "step": 54000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8176083264308964e-05,
+      "loss": 0.8154,
+      "step": 54500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8159350083247578e-05,
+      "loss": 0.7769,
+      "step": 55000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.814261690218619e-05,
+      "loss": 0.7822,
+      "step": 55500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8125883721124805e-05,
+      "loss": 0.7809,
+      "step": 56000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.810915054006342e-05,
+      "loss": 0.7934,
+      "step": 56500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8092417359002036e-05,
+      "loss": 0.7816,
+      "step": 57000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.807568417794065e-05,
+      "loss": 0.805,
+      "step": 57500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8058950996879263e-05,
+      "loss": 0.7964,
+      "step": 58000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8042217815817877e-05,
+      "loss": 0.766,
+      "step": 58500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.802548463475649e-05,
+      "loss": 0.7976,
+      "step": 59000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8008751453695104e-05,
+      "loss": 0.7881,
+      "step": 59500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.799201827263372e-05,
+      "loss": 0.7924,
+      "step": 60000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7975285091572335e-05,
+      "loss": 0.7863,
+      "step": 60500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.795855191051095e-05,
+      "loss": 0.7719,
+      "step": 61000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7941818729449562e-05,
+      "loss": 0.7795,
+      "step": 61500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7925085548388176e-05,
+      "loss": 0.7613,
+      "step": 62000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.790835236732679e-05,
+      "loss": 0.777,
+      "step": 62500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7891619186265407e-05,
+      "loss": 0.8009,
+      "step": 63000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.787488600520402e-05,
+      "loss": 0.7831,
+      "step": 63500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7858152824142637e-05,
+      "loss": 0.7811,
+      "step": 64000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.784141964308125e-05,
+      "loss": 0.7799,
+      "step": 64500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7824686462019865e-05,
+      "loss": 0.7823,
+      "step": 65000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7807953280958478e-05,
+      "loss": 0.777,
+      "step": 65500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7791220099897092e-05,
+      "loss": 0.774,
+      "step": 66000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7774486918835706e-05,
+      "loss": 0.7645,
+      "step": 66500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7757753737774323e-05,
+      "loss": 0.7481,
+      "step": 67000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7741020556712936e-05,
+      "loss": 0.7877,
+      "step": 67500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.772428737565155e-05,
+      "loss": 0.7734,
+      "step": 68000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7707554194590164e-05,
+      "loss": 0.8006,
+      "step": 68500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7690821013528777e-05,
+      "loss": 0.7815,
+      "step": 69000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.767408783246739e-05,
+      "loss": 0.779,
+      "step": 69500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7657354651406008e-05,
+      "loss": 0.7793,
+      "step": 70000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.764062147034462e-05,
+      "loss": 0.781,
+      "step": 70500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7623888289283235e-05,
+      "loss": 0.7621,
+      "step": 71000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.760715510822185e-05,
+      "loss": 0.7678,
+      "step": 71500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7590421927160463e-05,
+      "loss": 0.7682,
+      "step": 72000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.757368874609908e-05,
+      "loss": 0.7401,
+      "step": 72500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7556955565037693e-05,
+      "loss": 0.7831,
+      "step": 73000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7540222383976307e-05,
+      "loss": 0.772,
+      "step": 73500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.752348920291492e-05,
+      "loss": 0.7894,
+      "step": 74000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.7506756021853534e-05,
+      "loss": 0.7868,
+      "step": 74500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.749002284079215e-05,
+      "loss": 0.782,
+      "step": 75000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7473289659730765e-05,
+      "loss": 0.7693,
+      "step": 75500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.745655647866938e-05,
+      "loss": 0.7721,
+      "step": 76000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7439823297607996e-05,
+      "loss": 0.7713,
+      "step": 76500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.742309011654661e-05,
+      "loss": 0.749,
+      "step": 77000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7406356935485223e-05,
+      "loss": 0.7875,
+      "step": 77500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7389623754423837e-05,
+      "loss": 0.7601,
+      "step": 78000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.737289057336245e-05,
+      "loss": 0.7581,
+      "step": 78500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7356157392301064e-05,
+      "loss": 0.8048,
+      "step": 79000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.733942421123968e-05,
+      "loss": 0.7816,
+      "step": 79500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7322691030178295e-05,
+      "loss": 0.7828,
+      "step": 80000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7305957849116908e-05,
+      "loss": 0.7588,
+      "step": 80500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7289224668055522e-05,
+      "loss": 0.7496,
+      "step": 81000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7272491486994136e-05,
+      "loss": 0.7608,
+      "step": 81500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.725575830593275e-05,
+      "loss": 0.7753,
+      "step": 82000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7239025124871366e-05,
+      "loss": 0.7797,
+      "step": 82500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.722229194380998e-05,
+      "loss": 0.7341,
+      "step": 83000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7205558762748594e-05,
+      "loss": 0.8016,
+      "step": 83500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7188825581687207e-05,
+      "loss": 0.786,
+      "step": 84000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.717209240062582e-05,
+      "loss": 0.7833,
+      "step": 84500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7155359219564434e-05,
+      "loss": 0.7744,
+      "step": 85000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.713862603850305e-05,
+      "loss": 0.7916,
+      "step": 85500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7121892857441665e-05,
+      "loss": 0.7551,
+      "step": 86000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7105159676380282e-05,
+      "loss": 0.7826,
+      "step": 86500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7088426495318896e-05,
+      "loss": 0.7793,
+      "step": 87000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.707169331425751e-05,
+      "loss": 0.734,
+      "step": 87500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7054960133196123e-05,
+      "loss": 0.755,
+      "step": 88000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7038226952134737e-05,
+      "loss": 0.7508,
+      "step": 88500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.702149377107335e-05,
+      "loss": 0.7752,
+      "step": 89000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7004760590011968e-05,
+      "loss": 0.751,
+      "step": 89500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.698802740895058e-05,
+      "loss": 0.7251,
+      "step": 90000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6971294227889195e-05,
+      "loss": 0.7825,
+      "step": 90500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.695456104682781e-05,
+      "loss": 0.7629,
+      "step": 91000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6937827865766422e-05,
+      "loss": 0.7687,
+      "step": 91500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6921094684705036e-05,
+      "loss": 0.7788,
+      "step": 92000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6904361503643653e-05,
+      "loss": 0.774,
+      "step": 92500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6887628322582266e-05,
+      "loss": 0.768,
+      "step": 93000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.687089514152088e-05,
+      "loss": 0.7384,
+      "step": 93500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6854161960459494e-05,
+      "loss": 0.7617,
+      "step": 94000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6837428779398107e-05,
+      "loss": 0.7596,
+      "step": 94500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.682069559833672e-05,
+      "loss": 0.7665,
+      "step": 95000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6803962417275338e-05,
+      "loss": 0.7691,
+      "step": 95500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6787229236213952e-05,
+      "loss": 0.7761,
+      "step": 96000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6770496055152565e-05,
+      "loss": 0.7482,
+      "step": 96500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.675376287409118e-05,
+      "loss": 0.7573,
+      "step": 97000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6737029693029793e-05,
+      "loss": 0.7419,
+      "step": 97500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.672029651196841e-05,
+      "loss": 0.7379,
+      "step": 98000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.6703563330907023e-05,
+      "loss": 0.7459,
+      "step": 98500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6686830149845637e-05,
+      "loss": 0.7568,
+      "step": 99000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6670096968784254e-05,
+      "loss": 0.7476,
+      "step": 99500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6653363787722868e-05,
+      "loss": 0.7591,
+      "step": 100000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.663663060666148e-05,
+      "loss": 0.7517,
+      "step": 100500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6619897425600095e-05,
+      "loss": 0.7586,
+      "step": 101000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.660316424453871e-05,
+      "loss": 0.7695,
+      "step": 101500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6586431063477322e-05,
+      "loss": 0.7468,
+      "step": 102000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.656969788241594e-05,
+      "loss": 0.7477,
+      "step": 102500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6552964701354553e-05,
+      "loss": 0.75,
+      "step": 103000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6536231520293167e-05,
+      "loss": 0.7501,
+      "step": 103500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.651949833923178e-05,
+      "loss": 0.7283,
+      "step": 104000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6502765158170394e-05,
+      "loss": 0.7459,
+      "step": 104500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6486031977109008e-05,
+      "loss": 0.7795,
+      "step": 105000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6469298796047625e-05,
+      "loss": 0.7374,
+      "step": 105500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.645256561498624e-05,
+      "loss": 0.7455,
+      "step": 106000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6435832433924852e-05,
+      "loss": 0.7341,
+      "step": 106500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6419099252863466e-05,
+      "loss": 0.7746,
+      "step": 107000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.640236607180208e-05,
+      "loss": 0.7476,
+      "step": 107500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6385632890740696e-05,
+      "loss": 0.7603,
+      "step": 108000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.636889970967931e-05,
+      "loss": 0.7259,
+      "step": 108500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6352166528617924e-05,
+      "loss": 0.7625,
+      "step": 109000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.633543334755654e-05,
+      "loss": 0.757,
+      "step": 109500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6318700166495154e-05,
+      "loss": 0.7515,
+      "step": 110000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6301966985433768e-05,
+      "loss": 0.7404,
+      "step": 110500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6285233804372382e-05,
+      "loss": 0.7367,
+      "step": 111000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6268500623310995e-05,
+      "loss": 0.7493,
+      "step": 111500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6251767442249612e-05,
+      "loss": 0.7862,
+      "step": 112000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6235034261188226e-05,
+      "loss": 0.7553,
+      "step": 112500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.621830108012684e-05,
+      "loss": 0.7514,
+      "step": 113000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6201567899065453e-05,
+      "loss": 0.7365,
+      "step": 113500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6184834718004067e-05,
+      "loss": 0.7554,
+      "step": 114000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.616810153694268e-05,
+      "loss": 0.747,
+      "step": 114500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6151368355881298e-05,
+      "loss": 0.7808,
+      "step": 115000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.613463517481991e-05,
+      "loss": 0.7636,
+      "step": 115500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6117901993758525e-05,
+      "loss": 0.7655,
+      "step": 116000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.610116881269714e-05,
+      "loss": 0.7438,
+      "step": 116500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6084435631635752e-05,
+      "loss": 0.7249,
+      "step": 117000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6067702450574366e-05,
+      "loss": 0.7208,
+      "step": 117500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6050969269512983e-05,
+      "loss": 0.7463,
+      "step": 118000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6034236088451597e-05,
+      "loss": 0.7721,
+      "step": 118500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.601750290739021e-05,
+      "loss": 0.7317,
+      "step": 119000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6000769726328824e-05,
+      "loss": 0.7651,
+      "step": 119500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.5984036545267438e-05,
+      "loss": 0.7415,
+      "step": 120000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.596730336420605e-05,
+      "loss": 0.7797,
+      "step": 120500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.595057018314467e-05,
+      "loss": 0.7275,
+      "step": 121000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.5933837002083282e-05,
+      "loss": 0.7494,
+      "step": 121500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.59171038210219e-05,
+      "loss": 0.7651,
+      "step": 122000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.5900370639960513e-05,
+      "loss": 0.767,
+      "step": 122500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5883637458899126e-05,
+      "loss": 0.733,
+      "step": 123000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.586690427783774e-05,
+      "loss": 0.7648,
+      "step": 123500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5850171096776354e-05,
+      "loss": 0.7633,
+      "step": 124000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5833437915714967e-05,
+      "loss": 0.7493,
+      "step": 124500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5816704734653584e-05,
+      "loss": 0.7595,
+      "step": 125000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5799971553592198e-05,
+      "loss": 0.7527,
+      "step": 125500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.578323837253081e-05,
+      "loss": 0.7656,
+      "step": 126000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5766505191469425e-05,
+      "loss": 0.7534,
+      "step": 126500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.574977201040804e-05,
+      "loss": 0.7627,
+      "step": 127000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5733038829346653e-05,
+      "loss": 0.7498,
+      "step": 127500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.571630564828527e-05,
+      "loss": 0.7452,
+      "step": 128000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5699572467223883e-05,
+      "loss": 0.7844,
+      "step": 128500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5682839286162497e-05,
+      "loss": 0.7487,
+      "step": 129000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.566610610510111e-05,
+      "loss": 0.7739,
+      "step": 129500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5649372924039724e-05,
+      "loss": 0.746,
+      "step": 130000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5632639742978338e-05,
+      "loss": 0.7392,
+      "step": 130500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5615906561916955e-05,
+      "loss": 0.7729,
+      "step": 131000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.559917338085557e-05,
+      "loss": 0.7188,
+      "step": 131500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5582440199794182e-05,
+      "loss": 0.7491,
+      "step": 132000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.55657070187328e-05,
+      "loss": 0.7168,
+      "step": 132500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5548973837671413e-05,
+      "loss": 0.7247,
+      "step": 133000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5532240656610027e-05,
+      "loss": 0.7478,
+      "step": 133500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.551550747554864e-05,
+      "loss": 0.7459,
+      "step": 134000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5498774294487254e-05,
+      "loss": 0.7592,
+      "step": 134500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.548204111342587e-05,
+      "loss": 0.7476,
+      "step": 135000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5465307932364485e-05,
+      "loss": 0.7305,
+      "step": 135500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5448574751303098e-05,
+      "loss": 0.7461,
+      "step": 136000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5431841570241712e-05,
+      "loss": 0.7348,
+      "step": 136500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5415108389180326e-05,
+      "loss": 0.7517,
+      "step": 137000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.539837520811894e-05,
+      "loss": 0.7213,
+      "step": 137500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5381642027057556e-05,
+      "loss": 0.7519,
+      "step": 138000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.536490884599617e-05,
+      "loss": 0.7351,
+      "step": 138500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5348175664934784e-05,
+      "loss": 0.7488,
+      "step": 139000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5331442483873397e-05,
+      "loss": 0.7412,
+      "step": 139500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.531470930281201e-05,
+      "loss": 0.7274,
+      "step": 140000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5297976121750624e-05,
+      "loss": 0.7401,
+      "step": 140500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.528124294068924e-05,
+      "loss": 0.7388,
+      "step": 141000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5264509759627855e-05,
+      "loss": 0.7499,
+      "step": 141500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5247776578566469e-05,
+      "loss": 0.7574,
+      "step": 142000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5231043397505082e-05,
+      "loss": 0.7401,
+      "step": 142500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5214310216443698e-05,
+      "loss": 0.7302,
+      "step": 143000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5197577035382311e-05,
+      "loss": 0.7445,
+      "step": 143500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5180843854320929e-05,
+      "loss": 0.7283,
+      "step": 144000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5164110673259542e-05,
+      "loss": 0.7483,
+      "step": 144500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5147377492198156e-05,
+      "loss": 0.7442,
+      "step": 145000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5130644311136771e-05,
+      "loss": 0.7474,
+      "step": 145500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5113911130075385e-05,
+      "loss": 0.761,
+      "step": 146000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5097177949013998e-05,
+      "loss": 0.7299,
+      "step": 146500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5080444767952614e-05,
+      "loss": 0.7224,
+      "step": 147000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5063711586891227e-05,
+      "loss": 0.7541,
+      "step": 147500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5046978405829841e-05,
+      "loss": 0.7356,
+      "step": 148000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5030245224768456e-05,
+      "loss": 0.7366,
+      "step": 148500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.501351204370707e-05,
+      "loss": 0.7426,
+      "step": 149000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4996778862645684e-05,
+      "loss": 0.7556,
+      "step": 149500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4980045681584299e-05,
+      "loss": 0.728,
+      "step": 150000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4963312500522913e-05,
+      "loss": 0.7549,
+      "step": 150500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4946579319461526e-05,
+      "loss": 0.7371,
+      "step": 151000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4929846138400142e-05,
+      "loss": 0.747,
+      "step": 151500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.4913112957338755e-05,
+      "loss": 0.7535,
+      "step": 152000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4896379776277369e-05,
+      "loss": 0.7539,
+      "step": 152500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4879646595215984e-05,
+      "loss": 0.7588,
+      "step": 153000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4862913414154598e-05,
+      "loss": 0.7527,
+      "step": 153500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4846180233093212e-05,
+      "loss": 0.7229,
+      "step": 154000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4829447052031827e-05,
+      "loss": 0.7323,
+      "step": 154500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.481271387097044e-05,
+      "loss": 0.7106,
+      "step": 155000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4795980689909058e-05,
+      "loss": 0.7385,
+      "step": 155500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4779247508847671e-05,
+      "loss": 0.7382,
+      "step": 156000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4762514327786285e-05,
+      "loss": 0.7317,
+      "step": 156500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.47457811467249e-05,
+      "loss": 0.7467,
+      "step": 157000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4729047965663514e-05,
+      "loss": 0.7411,
+      "step": 157500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.4712314784602128e-05,
+      "loss": 0.749,
+      "step": 158000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4695581603540743e-05,
+      "loss": 0.7425,
+      "step": 158500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4678848422479357e-05,
+      "loss": 0.7585,
+      "step": 159000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4662115241417972e-05,
+      "loss": 0.7192,
+      "step": 159500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4645382060356586e-05,
+      "loss": 0.7609,
+      "step": 160000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.46286488792952e-05,
+      "loss": 0.7476,
+      "step": 160500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4611915698233815e-05,
+      "loss": 0.7186,
+      "step": 161000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4595182517172428e-05,
+      "loss": 0.7707,
+      "step": 161500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4578449336111042e-05,
+      "loss": 0.7271,
+      "step": 162000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4561716155049657e-05,
+      "loss": 0.7287,
+      "step": 162500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4544982973988271e-05,
+      "loss": 0.7343,
+      "step": 163000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4528249792926885e-05,
+      "loss": 0.7384,
+      "step": 163500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.45115166118655e-05,
+      "loss": 0.7428,
+      "step": 164000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4494783430804114e-05,
+      "loss": 0.701,
+      "step": 164500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4478050249742727e-05,
+      "loss": 0.7418,
+      "step": 165000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4461317068681343e-05,
+      "loss": 0.7228,
+      "step": 165500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4444583887619956e-05,
+      "loss": 0.7283,
+      "step": 166000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.442785070655857e-05,
+      "loss": 0.7504,
+      "step": 166500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4411117525497187e-05,
+      "loss": 0.7348,
+      "step": 167000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.43943843444358e-05,
+      "loss": 0.7355,
+      "step": 167500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4377651163374416e-05,
+      "loss": 0.7491,
+      "step": 168000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.436091798231303e-05,
+      "loss": 0.7449,
+      "step": 168500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4344184801251643e-05,
+      "loss": 0.7412,
+      "step": 169000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4327451620190259e-05,
+      "loss": 0.7242,
+      "step": 169500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4310718439128872e-05,
+      "loss": 0.7461,
+      "step": 170000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4293985258067486e-05,
+      "loss": 0.7472,
+      "step": 170500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4277252077006101e-05,
+      "loss": 0.726,
+      "step": 171000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4260518895944715e-05,
+      "loss": 0.7235,
+      "step": 171500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4243785714883329e-05,
+      "loss": 0.7188,
+      "step": 172000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4227052533821944e-05,
+      "loss": 0.746,
+      "step": 172500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4210319352760558e-05,
+      "loss": 0.7413,
+      "step": 173000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4193586171699171e-05,
+      "loss": 0.7224,
+      "step": 173500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4176852990637787e-05,
+      "loss": 0.7278,
+      "step": 174000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.41601198095764e-05,
+      "loss": 0.7645,
+      "step": 174500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4143386628515014e-05,
+      "loss": 0.736,
+      "step": 175000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.412665344745363e-05,
+      "loss": 0.745,
+      "step": 175500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4109920266392243e-05,
+      "loss": 0.7646,
+      "step": 176000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4093187085330857e-05,
+      "loss": 0.7386,
+      "step": 176500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4076453904269472e-05,
+      "loss": 0.7296,
+      "step": 177000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4059720723208086e-05,
+      "loss": 0.7251,
+      "step": 177500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.40429875421467e-05,
+      "loss": 0.729,
+      "step": 178000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4026254361085316e-05,
+      "loss": 0.7336,
+      "step": 178500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.400952118002393e-05,
+      "loss": 0.7301,
+      "step": 179000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3992787998962545e-05,
+      "loss": 0.7523,
+      "step": 179500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3976054817901159e-05,
+      "loss": 0.7283,
+      "step": 180000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3959321636839773e-05,
+      "loss": 0.7319,
+      "step": 180500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3942588455778388e-05,
+      "loss": 0.703,
+      "step": 181000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3925855274717002e-05,
+      "loss": 0.7678,
+      "step": 181500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3909122093655615e-05,
+      "loss": 0.7205,
+      "step": 182000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.389238891259423e-05,
+      "loss": 0.7314,
+      "step": 182500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3875655731532844e-05,
+      "loss": 0.759,
+      "step": 183000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3858922550471458e-05,
+      "loss": 0.7179,
+      "step": 183500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3842189369410073e-05,
+      "loss": 0.7198,
+      "step": 184000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3825456188348687e-05,
+      "loss": 0.7467,
+      "step": 184500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.38087230072873e-05,
+      "loss": 0.7274,
+      "step": 185000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3791989826225916e-05,
+      "loss": 0.719,
+      "step": 185500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.377525664516453e-05,
+      "loss": 0.7258,
+      "step": 186000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3758523464103143e-05,
+      "loss": 0.7299,
+      "step": 186500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3741790283041759e-05,
+      "loss": 0.7457,
+      "step": 187000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3725057101980372e-05,
+      "loss": 0.733,
+      "step": 187500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3708323920918986e-05,
+      "loss": 0.7276,
+      "step": 188000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3691590739857601e-05,
+      "loss": 0.7431,
+      "step": 188500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3674857558796215e-05,
+      "loss": 0.7174,
+      "step": 189000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3658124377734829e-05,
+      "loss": 0.7217,
+      "step": 189500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3641391196673446e-05,
+      "loss": 0.7253,
+      "step": 190000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.362465801561206e-05,
+      "loss": 0.7128,
+      "step": 190500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3607924834550675e-05,
+      "loss": 0.7191,
+      "step": 191000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3591191653489288e-05,
+      "loss": 0.734,
+      "step": 191500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3574458472427902e-05,
+      "loss": 0.7225,
+      "step": 192000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3557725291366517e-05,
+      "loss": 0.7324,
+      "step": 192500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3540992110305131e-05,
+      "loss": 0.742,
+      "step": 193000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3524258929243745e-05,
+      "loss": 0.7072,
+      "step": 193500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.350752574818236e-05,
+      "loss": 0.7196,
+      "step": 194000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3490792567120974e-05,
+      "loss": 0.7453,
+      "step": 194500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3474059386059589e-05,
+      "loss": 0.7397,
+      "step": 195000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3457326204998203e-05,
+      "loss": 0.7303,
+      "step": 195500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3440593023936816e-05,
+      "loss": 0.7476,
+      "step": 196000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3423859842875432e-05,
+      "loss": 0.7371,
+      "step": 196500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3407126661814045e-05,
+      "loss": 0.7124,
+      "step": 197000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3390393480752659e-05,
+      "loss": 0.741,
+      "step": 197500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3373660299691274e-05,
+      "loss": 0.7117,
+      "step": 198000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3356927118629888e-05,
+      "loss": 0.7492,
+      "step": 198500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3340193937568502e-05,
+      "loss": 0.7412,
+      "step": 199000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3323460756507117e-05,
+      "loss": 0.7121,
+      "step": 199500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.330672757544573e-05,
+      "loss": 0.7273,
+      "step": 200000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3289994394384344e-05,
+      "loss": 0.717,
+      "step": 200500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.327326121332296e-05,
+      "loss": 0.7306,
+      "step": 201000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3256528032261575e-05,
+      "loss": 0.7453,
+      "step": 201500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.323979485120019e-05,
+      "loss": 0.7462,
+      "step": 202000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3223061670138804e-05,
+      "loss": 0.7039,
+      "step": 202500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3206328489077418e-05,
+      "loss": 0.7303,
+      "step": 203000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3189595308016033e-05,
+      "loss": 0.7149,
+      "step": 203500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3172862126954647e-05,
+      "loss": 0.7215,
+      "step": 204000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.315612894589326e-05,
+      "loss": 0.7124,
+      "step": 204500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3139395764831876e-05,
+      "loss": 0.7188,
+      "step": 205000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.312266258377049e-05,
+      "loss": 0.7316,
+      "step": 205500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3105929402709103e-05,
+      "loss": 0.7301,
+      "step": 206000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3089196221647718e-05,
+      "loss": 0.7231,
+      "step": 206500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3072463040586332e-05,
+      "loss": 0.7476,
+      "step": 207000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3055729859524945e-05,
+      "loss": 0.7326,
+      "step": 207500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.303899667846356e-05,
+      "loss": 0.7103,
+      "step": 208000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3022263497402174e-05,
+      "loss": 0.7394,
+      "step": 208500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3005530316340788e-05,
+      "loss": 0.7321,
+      "step": 209000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2988797135279403e-05,
+      "loss": 0.7124,
+      "step": 209500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2972063954218017e-05,
+      "loss": 0.7368,
+      "step": 210000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.295533077315663e-05,
+      "loss": 0.7256,
+      "step": 210500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2938597592095246e-05,
+      "loss": 0.7193,
+      "step": 211000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.292186441103386e-05,
+      "loss": 0.7524,
+      "step": 211500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2905131229972473e-05,
+      "loss": 0.7529,
+      "step": 212000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2888398048911089e-05,
+      "loss": 0.7294,
+      "step": 212500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2871664867849704e-05,
+      "loss": 0.717,
+      "step": 213000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.285493168678832e-05,
+      "loss": 0.7142,
+      "step": 213500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2838198505726933e-05,
+      "loss": 0.7115,
+      "step": 214000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2821465324665547e-05,
+      "loss": 0.7417,
+      "step": 214500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2804732143604162e-05,
+      "loss": 0.7258,
+      "step": 215000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2787998962542776e-05,
+      "loss": 0.7437,
+      "step": 215500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.277126578148139e-05,
+      "loss": 0.7191,
+      "step": 216000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2754532600420005e-05,
+      "loss": 0.7223,
+      "step": 216500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2737799419358618e-05,
+      "loss": 0.7273,
+      "step": 217000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2721066238297232e-05,
+      "loss": 0.7169,
+      "step": 217500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.2704333057235847e-05,
+      "loss": 0.7136,
+      "step": 218000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2687599876174461e-05,
+      "loss": 0.7154,
+      "step": 218500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2670866695113075e-05,
+      "loss": 0.7153,
+      "step": 219000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.265413351405169e-05,
+      "loss": 0.7416,
+      "step": 219500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2637400332990304e-05,
+      "loss": 0.7129,
+      "step": 220000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2620667151928917e-05,
+      "loss": 0.7042,
+      "step": 220500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2603933970867533e-05,
+      "loss": 0.7512,
+      "step": 221000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2587200789806146e-05,
+      "loss": 0.7154,
+      "step": 221500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.257046760874476e-05,
+      "loss": 0.7301,
+      "step": 222000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2553734427683375e-05,
+      "loss": 0.7171,
+      "step": 222500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2537001246621989e-05,
+      "loss": 0.7339,
+      "step": 223000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2520268065560603e-05,
+      "loss": 0.7228,
+      "step": 223500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2503534884499218e-05,
+      "loss": 0.7122,
+      "step": 224000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2486801703437833e-05,
+      "loss": 0.7348,
+      "step": 224500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2470068522376449e-05,
+      "loss": 0.7422,
+      "step": 225000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2453335341315062e-05,
+      "loss": 0.704,
+      "step": 225500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2436602160253676e-05,
+      "loss": 0.7235,
+      "step": 226000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2419868979192291e-05,
+      "loss": 0.7601,
+      "step": 226500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2403135798130905e-05,
+      "loss": 0.7139,
+      "step": 227000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2386402617069519e-05,
+      "loss": 0.7178,
+      "step": 227500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2369669436008134e-05,
+      "loss": 0.7456,
+      "step": 228000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2352936254946748e-05,
+      "loss": 0.7082,
+      "step": 228500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2336203073885361e-05,
+      "loss": 0.7348,
+      "step": 229000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2319469892823977e-05,
+      "loss": 0.7037,
+      "step": 229500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.230273671176259e-05,
+      "loss": 0.7174,
+      "step": 230000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2286003530701206e-05,
+      "loss": 0.7219,
+      "step": 230500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.226927034963982e-05,
+      "loss": 0.7153,
+      "step": 231000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2252537168578433e-05,
+      "loss": 0.7515,
+      "step": 231500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2235803987517048e-05,
+      "loss": 0.7271,
+      "step": 232000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2219070806455662e-05,
+      "loss": 0.7004,
+      "step": 232500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2202337625394276e-05,
+      "loss": 0.71,
+      "step": 233000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2185604444332891e-05,
+      "loss": 0.7032,
+      "step": 233500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2168871263271505e-05,
+      "loss": 0.725,
+      "step": 234000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2152138082210118e-05,
+      "loss": 0.7063,
+      "step": 234500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2135404901148734e-05,
+      "loss": 0.7033,
+      "step": 235000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2118671720087347e-05,
+      "loss": 0.7501,
+      "step": 235500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2101938539025964e-05,
+      "loss": 0.7052,
+      "step": 236000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2085205357964578e-05,
+      "loss": 0.7043,
+      "step": 236500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2068472176903192e-05,
+      "loss": 0.7358,
+      "step": 237000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2051738995841807e-05,
+      "loss": 0.7252,
+      "step": 237500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.203500581478042e-05,
+      "loss": 0.7326,
+      "step": 238000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2018272633719034e-05,
+      "loss": 0.722,
+      "step": 238500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.200153945265765e-05,
+      "loss": 0.7056,
+      "step": 239000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1984806271596263e-05,
+      "loss": 0.715,
+      "step": 239500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1968073090534877e-05,
+      "loss": 0.7174,
+      "step": 240000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1951339909473492e-05,
+      "loss": 0.7293,
+      "step": 240500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1934606728412106e-05,
+      "loss": 0.7376,
+      "step": 241000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.191787354735072e-05,
+      "loss": 0.7412,
+      "step": 241500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.1901140366289335e-05,
+      "loss": 0.6738,
+      "step": 242000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1884407185227949e-05,
+      "loss": 0.7222,
+      "step": 242500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1867674004166562e-05,
+      "loss": 0.7344,
+      "step": 243000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1850940823105178e-05,
+      "loss": 0.7292,
+      "step": 243500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1834207642043791e-05,
+      "loss": 0.7157,
+      "step": 244000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1817474460982405e-05,
+      "loss": 0.7382,
+      "step": 244500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.180074127992102e-05,
+      "loss": 0.7161,
+      "step": 245000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1784008098859634e-05,
+      "loss": 0.7196,
+      "step": 245500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1767274917798248e-05,
+      "loss": 0.7207,
+      "step": 246000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1750541736736863e-05,
+      "loss": 0.7081,
+      "step": 246500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1733808555675477e-05,
+      "loss": 0.6827,
+      "step": 247000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1717075374614094e-05,
+      "loss": 0.7289,
+      "step": 247500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.1700342193552707e-05,
+      "loss": 0.7434,
+      "step": 248000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1683609012491321e-05,
+      "loss": 0.717,
+      "step": 248500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1666875831429936e-05,
+      "loss": 0.734,
+      "step": 249000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.165014265036855e-05,
+      "loss": 0.7083,
+      "step": 249500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1633409469307164e-05,
+      "loss": 0.7315,
+      "step": 250000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1616676288245779e-05,
+      "loss": 0.7101,
+      "step": 250500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1599943107184393e-05,
+      "loss": 0.7059,
+      "step": 251000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1583209926123006e-05,
+      "loss": 0.7199,
+      "step": 251500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1566476745061622e-05,
+      "loss": 0.7157,
+      "step": 252000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1549743564000235e-05,
+      "loss": 0.7188,
+      "step": 252500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1533010382938849e-05,
+      "loss": 0.714,
+      "step": 253000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1516277201877464e-05,
+      "loss": 0.7351,
+      "step": 253500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1499544020816078e-05,
+      "loss": 0.6861,
+      "step": 254000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1482810839754692e-05,
+      "loss": 0.7069,
+      "step": 254500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1466077658693307e-05,
+      "loss": 0.7163,
+      "step": 255000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.144934447763192e-05,
+      "loss": 0.7343,
+      "step": 255500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1432611296570534e-05,
+      "loss": 0.7085,
+      "step": 256000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.141587811550915e-05,
+      "loss": 0.7281,
+      "step": 256500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1399144934447763e-05,
+      "loss": 0.7217,
+      "step": 257000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1382411753386377e-05,
+      "loss": 0.7433,
+      "step": 257500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1365678572324992e-05,
+      "loss": 0.752,
+      "step": 258000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1348945391263606e-05,
+      "loss": 0.7236,
+      "step": 258500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1332212210202223e-05,
+      "loss": 0.6982,
+      "step": 259000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1315479029140837e-05,
+      "loss": 0.7155,
+      "step": 259500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.129874584807945e-05,
+      "loss": 0.7229,
+      "step": 260000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1282012667018066e-05,
+      "loss": 0.7252,
+      "step": 260500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.126527948595668e-05,
+      "loss": 0.7049,
+      "step": 261000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1248546304895293e-05,
+      "loss": 0.6975,
+      "step": 261500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1231813123833908e-05,
+      "loss": 0.7088,
+      "step": 262000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1215079942772522e-05,
+      "loss": 0.7056,
+      "step": 262500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1198346761711136e-05,
+      "loss": 0.713,
+      "step": 263000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1181613580649751e-05,
+      "loss": 0.7069,
+      "step": 263500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1164880399588365e-05,
+      "loss": 0.7027,
+      "step": 264000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1148147218526978e-05,
+      "loss": 0.7247,
+      "step": 264500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1131414037465594e-05,
+      "loss": 0.7246,
+      "step": 265000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1114680856404207e-05,
+      "loss": 0.7169,
+      "step": 265500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1097947675342823e-05,
+      "loss": 0.7247,
+      "step": 266000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1081214494281436e-05,
+      "loss": 0.7131,
+      "step": 266500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.106448131322005e-05,
+      "loss": 0.715,
+      "step": 267000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1047748132158665e-05,
+      "loss": 0.7352,
+      "step": 267500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1031014951097279e-05,
+      "loss": 0.7144,
+      "step": 268000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1014281770035892e-05,
+      "loss": 0.7232,
+      "step": 268500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0997548588974508e-05,
+      "loss": 0.6849,
+      "step": 269000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0980815407913121e-05,
+      "loss": 0.7161,
+      "step": 269500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0964082226851735e-05,
+      "loss": 0.7275,
+      "step": 270000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0947349045790352e-05,
+      "loss": 0.6719,
+      "step": 270500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0930615864728966e-05,
+      "loss": 0.708,
+      "step": 271000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.0913882683667581e-05,
+      "loss": 0.7125,
+      "step": 271500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0897149502606195e-05,
+      "loss": 0.7051,
+      "step": 272000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0880416321544808e-05,
+      "loss": 0.6957,
+      "step": 272500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0863683140483424e-05,
+      "loss": 0.7115,
+      "step": 273000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0846949959422037e-05,
+      "loss": 0.6966,
+      "step": 273500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0830216778360651e-05,
+      "loss": 0.703,
+      "step": 274000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0813483597299266e-05,
+      "loss": 0.7069,
+      "step": 274500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.079675041623788e-05,
+      "loss": 0.7038,
+      "step": 275000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0780017235176494e-05,
+      "loss": 0.6937,
+      "step": 275500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0763284054115109e-05,
+      "loss": 0.6794,
+      "step": 276000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0746550873053723e-05,
+      "loss": 0.6789,
+      "step": 276500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0729817691992336e-05,
+      "loss": 0.7016,
+      "step": 277000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.0713084510930952e-05,
+      "loss": 0.7232,
+      "step": 277500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0696351329869565e-05,
+      "loss": 0.7148,
+      "step": 278000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0679618148808179e-05,
+      "loss": 0.7023,
+      "step": 278500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0662884967746794e-05,
+      "loss": 0.7351,
+      "step": 279000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0646151786685408e-05,
+      "loss": 0.6924,
+      "step": 279500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0629418605624022e-05,
+      "loss": 0.728,
+      "step": 280000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0612685424562637e-05,
+      "loss": 0.7198,
+      "step": 280500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.059595224350125e-05,
+      "loss": 0.712,
+      "step": 281000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0579219062439864e-05,
+      "loss": 0.729,
+      "step": 281500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0562485881378481e-05,
+      "loss": 0.6932,
+      "step": 282000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0545752700317095e-05,
+      "loss": 0.7212,
+      "step": 282500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.052901951925571e-05,
+      "loss": 0.7447,
+      "step": 283000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0512286338194324e-05,
+      "loss": 0.7325,
+      "step": 283500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0495553157132938e-05,
+      "loss": 0.6866,
+      "step": 284000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0478819976071553e-05,
+      "loss": 0.707,
+      "step": 284500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0462086795010167e-05,
+      "loss": 0.7256,
+      "step": 285000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.044535361394878e-05,
+      "loss": 0.6986,
+      "step": 285500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0428620432887396e-05,
+      "loss": 0.7238,
+      "step": 286000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.041188725182601e-05,
+      "loss": 0.707,
+      "step": 286500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0395154070764623e-05,
+      "loss": 0.7119,
+      "step": 287000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0378420889703238e-05,
+      "loss": 0.7164,
+      "step": 287500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0361687708641852e-05,
+      "loss": 0.711,
+      "step": 288000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0344954527580466e-05,
+      "loss": 0.6988,
+      "step": 288500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0328221346519081e-05,
+      "loss": 0.6844,
+      "step": 289000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0311488165457695e-05,
+      "loss": 0.7007,
+      "step": 289500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0294754984396308e-05,
+      "loss": 0.6879,
+      "step": 290000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0278021803334924e-05,
+      "loss": 0.6787,
+      "step": 290500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0261288622273537e-05,
+      "loss": 0.7218,
+      "step": 291000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0244555441212151e-05,
+      "loss": 0.7189,
+      "step": 291500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0227822260150766e-05,
+      "loss": 0.7284,
+      "step": 292000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.021108907908938e-05,
+      "loss": 0.7056,
+      "step": 292500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0194355898027994e-05,
+      "loss": 0.721,
+      "step": 293000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.017762271696661e-05,
+      "loss": 0.7161,
+      "step": 293500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0160889535905224e-05,
+      "loss": 0.7093,
+      "step": 294000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.014415635484384e-05,
+      "loss": 0.6968,
+      "step": 294500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0127423173782453e-05,
+      "loss": 0.7075,
+      "step": 295000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0110689992721067e-05,
+      "loss": 0.7183,
+      "step": 295500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0093956811659682e-05,
+      "loss": 0.7218,
+      "step": 296000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0077223630598296e-05,
+      "loss": 0.7151,
+      "step": 296500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.006049044953691e-05,
+      "loss": 0.7078,
+      "step": 297000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0043757268475525e-05,
+      "loss": 0.7056,
+      "step": 297500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0027024087414139e-05,
+      "loss": 0.713,
+      "step": 298000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0010290906352752e-05,
+      "loss": 0.7068,
+      "step": 298500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.993557725291368e-06,
+      "loss": 0.6865,
+      "step": 299000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.976824544229981e-06,
+      "loss": 0.7117,
+      "step": 299500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.960091363168595e-06,
+      "loss": 0.6941,
+      "step": 300000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.94335818210721e-06,
+      "loss": 0.6784,
+      "step": 300500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.926625001045824e-06,
+      "loss": 0.7178,
+      "step": 301000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.90989181998444e-06,
+      "loss": 0.6891,
+      "step": 301500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.893158638923053e-06,
+      "loss": 0.7026,
+      "step": 302000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.876425457861668e-06,
+      "loss": 0.7193,
+      "step": 302500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.859692276800282e-06,
+      "loss": 0.7182,
+      "step": 303000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.842959095738897e-06,
+      "loss": 0.7037,
+      "step": 303500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.826225914677511e-06,
+      "loss": 0.7142,
+      "step": 304000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.809492733616125e-06,
+      "loss": 0.7172,
+      "step": 304500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.79275955255474e-06,
+      "loss": 0.6979,
+      "step": 305000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.776026371493354e-06,
+      "loss": 0.697,
+      "step": 305500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.759293190431967e-06,
+      "loss": 0.7162,
+      "step": 306000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.742560009370583e-06,
+      "loss": 0.7054,
+      "step": 306500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.725826828309196e-06,
+      "loss": 0.7073,
+      "step": 307000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.70909364724781e-06,
+      "loss": 0.7117,
+      "step": 307500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.692360466186425e-06,
+      "loss": 0.6998,
+      "step": 308000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.67562728512504e-06,
+      "loss": 0.7088,
+      "step": 308500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.658894104063654e-06,
+      "loss": 0.6948,
+      "step": 309000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.642160923002268e-06,
+      "loss": 0.7215,
+      "step": 309500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.625427741940883e-06,
+      "loss": 0.6936,
+      "step": 310000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.608694560879497e-06,
+      "loss": 0.7172,
+      "step": 310500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.59196137981811e-06,
+      "loss": 0.7098,
+      "step": 311000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.575228198756726e-06,
+      "loss": 0.731,
+      "step": 311500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.55849501769534e-06,
+      "loss": 0.7204,
+      "step": 312000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.541761836633953e-06,
+      "loss": 0.714,
+      "step": 312500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.525028655572569e-06,
+      "loss": 0.6965,
+      "step": 313000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.508295474511182e-06,
+      "loss": 0.7104,
+      "step": 313500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.491562293449798e-06,
+      "loss": 0.7244,
+      "step": 314000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.474829112388411e-06,
+      "loss": 0.6985,
+      "step": 314500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458095931327027e-06,
+      "loss": 0.7091,
+      "step": 315000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44136275026564e-06,
+      "loss": 0.7056,
+      "step": 315500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.424629569204254e-06,
+      "loss": 0.7207,
+      "step": 316000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40789638814287e-06,
+      "loss": 0.7346,
+      "step": 316500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.391163207081483e-06,
+      "loss": 0.6894,
+      "step": 317000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.374430026020097e-06,
+      "loss": 0.7028,
+      "step": 317500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357696844958712e-06,
+      "loss": 0.7048,
+      "step": 318000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.340963663897326e-06,
+      "loss": 0.7083,
+      "step": 318500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.32423048283594e-06,
+      "loss": 0.6895,
+      "step": 319000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.307497301774555e-06,
+      "loss": 0.7022,
+      "step": 319500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29076412071317e-06,
+      "loss": 0.7353,
+      "step": 320000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.274030939651784e-06,
+      "loss": 0.7418,
+      "step": 320500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.257297758590397e-06,
+      "loss": 0.6897,
+      "step": 321000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240564577529013e-06,
+      "loss": 0.6675,
+      "step": 321500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223831396467626e-06,
+      "loss": 0.6751,
+      "step": 322000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.20709821540624e-06,
+      "loss": 0.7157,
+      "step": 322500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.190365034344855e-06,
+      "loss": 0.72,
+      "step": 323000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173631853283469e-06,
+      "loss": 0.6988,
+      "step": 323500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156898672222083e-06,
+      "loss": 0.6982,
+      "step": 324000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140165491160698e-06,
+      "loss": 0.6965,
+      "step": 324500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.123432310099312e-06,
+      "loss": 0.6938,
+      "step": 325000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.106699129037927e-06,
+      "loss": 0.696,
+      "step": 325500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.08996594797654e-06,
+      "loss": 0.7088,
+      "step": 326000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.073232766915156e-06,
+      "loss": 0.7205,
+      "step": 326500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.05649958585377e-06,
+      "loss": 0.7055,
+      "step": 327000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.039766404792383e-06,
+      "loss": 0.6873,
+      "step": 327500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.023033223730999e-06,
+      "loss": 0.7048,
+      "step": 328000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006300042669612e-06,
+      "loss": 0.7041,
+      "step": 328500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.989566861608226e-06,
+      "loss": 0.7302,
+      "step": 329000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.972833680546841e-06,
+      "loss": 0.7097,
+      "step": 329500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.956100499485455e-06,
+      "loss": 0.6946,
+      "step": 330000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939367318424068e-06,
+      "loss": 0.7361,
+      "step": 330500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922634137362684e-06,
+      "loss": 0.6961,
+      "step": 331000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9059009563013e-06,
+      "loss": 0.7008,
+      "step": 331500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.889167775239913e-06,
+      "loss": 0.7024,
+      "step": 332000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.872434594178526e-06,
+      "loss": 0.7028,
+      "step": 332500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.855701413117142e-06,
+      "loss": 0.713,
+      "step": 333000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.838968232055755e-06,
+      "loss": 0.6934,
+      "step": 333500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.822235050994369e-06,
+      "loss": 0.6979,
+      "step": 334000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.805501869932984e-06,
+      "loss": 0.699,
+      "step": 334500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.788768688871598e-06,
+      "loss": 0.7208,
+      "step": 335000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.772035507810213e-06,
+      "loss": 0.7121,
+      "step": 335500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.755302326748827e-06,
+      "loss": 0.7095,
+      "step": 336000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.73856914568744e-06,
+      "loss": 0.7095,
+      "step": 336500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.721835964626056e-06,
+      "loss": 0.7052,
+      "step": 337000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.705102783564671e-06,
+      "loss": 0.6974,
+      "step": 337500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.688369602503285e-06,
+      "loss": 0.6845,
+      "step": 338000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.671636421441899e-06,
+      "loss": 0.7137,
+      "step": 338500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.654903240380514e-06,
+      "loss": 0.6969,
+      "step": 339000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.638170059319128e-06,
+      "loss": 0.7083,
+      "step": 339500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.621436878257741e-06,
+      "loss": 0.6677,
+      "step": 340000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.604703697196357e-06,
+      "loss": 0.7216,
+      "step": 340500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.58797051613497e-06,
+      "loss": 0.7121,
+      "step": 341000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.571237335073584e-06,
+      "loss": 0.6989,
+      "step": 341500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.5545041540122e-06,
+      "loss": 0.7357,
+      "step": 342000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.537770972950815e-06,
+      "loss": 0.7041,
+      "step": 342500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.521037791889428e-06,
+      "loss": 0.7043,
+      "step": 343000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.504304610828042e-06,
+      "loss": 0.6784,
+      "step": 343500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.487571429766657e-06,
+      "loss": 0.6839,
+      "step": 344000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.470838248705271e-06,
+      "loss": 0.6995,
+      "step": 344500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.454105067643885e-06,
+      "loss": 0.7178,
+      "step": 345000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.4373718865825e-06,
+      "loss": 0.6731,
+      "step": 345500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.420638705521114e-06,
+      "loss": 0.685,
+      "step": 346000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.403905524459727e-06,
+      "loss": 0.6758,
+      "step": 346500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.387172343398343e-06,
+      "loss": 0.6718,
+      "step": 347000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.370439162336956e-06,
+      "loss": 0.6935,
+      "step": 347500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.35370598127557e-06,
+      "loss": 0.6986,
+      "step": 348000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.336972800214185e-06,
+      "loss": 0.6879,
+      "step": 348500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.3202396191528e-06,
+      "loss": 0.7093,
+      "step": 349000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.303506438091414e-06,
+      "loss": 0.6951,
+      "step": 349500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.286773257030028e-06,
+      "loss": 0.6957,
+      "step": 350000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.270040075968643e-06,
+      "loss": 0.6775,
+      "step": 350500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.253306894907257e-06,
+      "loss": 0.6856,
+      "step": 351000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.23657371384587e-06,
+      "loss": 0.6878,
+      "step": 351500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.219840532784486e-06,
+      "loss": 0.7127,
+      "step": 352000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.2031073517231e-06,
+      "loss": 0.6918,
+      "step": 352500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.186374170661713e-06,
+      "loss": 0.7071,
+      "step": 353000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.169640989600329e-06,
+      "loss": 0.7213,
+      "step": 353500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.152907808538944e-06,
+      "loss": 0.6969,
+      "step": 354000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.136174627477558e-06,
+      "loss": 0.6922,
+      "step": 354500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.119441446416171e-06,
+      "loss": 0.7182,
+      "step": 355000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.102708265354787e-06,
+      "loss": 0.7215,
+      "step": 355500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.0859750842934e-06,
+      "loss": 0.7216,
+      "step": 356000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.069241903232014e-06,
+      "loss": 0.693,
+      "step": 356500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.05250872217063e-06,
+      "loss": 0.71,
+      "step": 357000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.035775541109243e-06,
+      "loss": 0.7132,
+      "step": 357500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.019042360047857e-06,
+      "loss": 0.7119,
+      "step": 358000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.002309178986472e-06,
+      "loss": 0.7275,
+      "step": 358500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.985575997925086e-06,
+      "loss": 0.6845,
+      "step": 359000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.9688428168637e-06,
+      "loss": 0.7144,
+      "step": 359500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.952109635802315e-06,
+      "loss": 0.7127,
+      "step": 360000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.93537645474093e-06,
+      "loss": 0.727,
+      "step": 360500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.918643273679544e-06,
+      "loss": 0.6947,
+      "step": 361000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.901910092618157e-06,
+      "loss": 0.6901,
+      "step": 361500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.885176911556773e-06,
+      "loss": 0.6752,
+      "step": 362000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.868443730495386e-06,
+      "loss": 0.6827,
+      "step": 362500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.851710549434e-06,
+      "loss": 0.6785,
+      "step": 363000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.834977368372615e-06,
+      "loss": 0.6852,
+      "step": 363500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.818244187311229e-06,
+      "loss": 0.7021,
+      "step": 364000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.801511006249843e-06,
+      "loss": 0.6825,
+      "step": 364500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.784777825188458e-06,
+      "loss": 0.69,
+      "step": 365000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.768044644127073e-06,
+      "loss": 0.6904,
+      "step": 365500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.751311463065687e-06,
+      "loss": 0.6936,
+      "step": 366000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.7345782820043e-06,
+      "loss": 0.716,
+      "step": 366500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.717845100942916e-06,
+      "loss": 0.7102,
+      "step": 367000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.70111191988153e-06,
+      "loss": 0.697,
+      "step": 367500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.684378738820143e-06,
+      "loss": 0.7241,
+      "step": 368000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.667645557758759e-06,
+      "loss": 0.6712,
+      "step": 368500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.650912376697372e-06,
+      "loss": 0.6975,
+      "step": 369000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.634179195635986e-06,
+      "loss": 0.7205,
+      "step": 369500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.617446014574601e-06,
+      "loss": 0.6796,
+      "step": 370000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.600712833513215e-06,
+      "loss": 0.697,
+      "step": 370500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.5839796524518294e-06,
+      "loss": 0.6595,
+      "step": 371000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.567246471390445e-06,
+      "loss": 0.6935,
+      "step": 371500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.550513290329059e-06,
+      "loss": 0.683,
+      "step": 372000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.533780109267673e-06,
+      "loss": 0.6954,
+      "step": 372500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.5170469282062874e-06,
+      "loss": 0.672,
+      "step": 373000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.500313747144902e-06,
+      "loss": 0.6945,
+      "step": 373500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.483580566083516e-06,
+      "loss": 0.6856,
+      "step": 374000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.46684738502213e-06,
+      "loss": 0.678,
+      "step": 374500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.450114203960745e-06,
+      "loss": 0.6923,
+      "step": 375000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.433381022899358e-06,
+      "loss": 0.6811,
+      "step": 375500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.416647841837973e-06,
+      "loss": 0.7011,
+      "step": 376000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.399914660776587e-06,
+      "loss": 0.687,
+      "step": 376500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.383181479715203e-06,
+      "loss": 0.6837,
+      "step": 377000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.366448298653816e-06,
+      "loss": 0.6939,
+      "step": 377500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.349715117592431e-06,
+      "loss": 0.7025,
+      "step": 378000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.332981936531045e-06,
+      "loss": 0.6849,
+      "step": 378500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.316248755469659e-06,
+      "loss": 0.7049,
+      "step": 379000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.299515574408273e-06,
+      "loss": 0.6878,
+      "step": 379500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.282782393346888e-06,
+      "loss": 0.7159,
+      "step": 380000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.266049212285502e-06,
+      "loss": 0.6803,
+      "step": 380500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.249316031224116e-06,
+      "loss": 0.6796,
+      "step": 381000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.2325828501627305e-06,
+      "loss": 0.6971,
+      "step": 381500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.215849669101345e-06,
+      "loss": 0.6891,
+      "step": 382000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.199116488039959e-06,
+      "loss": 0.6886,
+      "step": 382500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.182383306978574e-06,
+      "loss": 0.703,
+      "step": 383000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.1656501259171885e-06,
+      "loss": 0.6894,
+      "step": 383500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.148916944855803e-06,
+      "loss": 0.7115,
+      "step": 384000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.132183763794417e-06,
+      "loss": 0.6946,
+      "step": 384500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.115450582733031e-06,
+      "loss": 0.7154,
+      "step": 385000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.098717401671646e-06,
+      "loss": 0.688,
+      "step": 385500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.081984220610259e-06,
+      "loss": 0.6977,
+      "step": 386000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.065251039548874e-06,
+      "loss": 0.6658,
+      "step": 386500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.048517858487488e-06,
+      "loss": 0.6964,
+      "step": 387000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.031784677426102e-06,
+      "loss": 0.7053,
+      "step": 387500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.0150514963647165e-06,
+      "loss": 0.6523,
+      "step": 388000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.998318315303332e-06,
+      "loss": 0.6794,
+      "step": 388500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.981585134241946e-06,
+      "loss": 0.6978,
+      "step": 389000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.96485195318056e-06,
+      "loss": 0.6705,
+      "step": 389500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.9481187721191745e-06,
+      "loss": 0.7086,
+      "step": 390000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.931385591057789e-06,
+      "loss": 0.6893,
+      "step": 390500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.914652409996403e-06,
+      "loss": 0.6755,
+      "step": 391000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.897919228935017e-06,
+      "loss": 0.6836,
+      "step": 391500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.881186047873632e-06,
+      "loss": 0.6981,
+      "step": 392000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.864452866812245e-06,
+      "loss": 0.6965,
+      "step": 392500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.84771968575086e-06,
+      "loss": 0.6845,
+      "step": 393000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.830986504689474e-06,
+      "loss": 0.6598,
+      "step": 393500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.814253323628088e-06,
+      "loss": 0.6998,
+      "step": 394000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.797520142566703e-06,
+      "loss": 0.698,
+      "step": 394500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.780786961505318e-06,
+      "loss": 0.6883,
+      "step": 395000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.764053780443932e-06,
+      "loss": 0.6845,
+      "step": 395500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.747320599382546e-06,
+      "loss": 0.6952,
+      "step": 396000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.7305874183211605e-06,
+      "loss": 0.6912,
+      "step": 396500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.713854237259775e-06,
+      "loss": 0.703,
+      "step": 397000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.6971210561983895e-06,
+      "loss": 0.687,
+      "step": 397500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.680387875137003e-06,
+      "loss": 0.6867,
+      "step": 398000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.663654694075618e-06,
+      "loss": 0.6989,
+      "step": 398500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.646921513014232e-06,
+      "loss": 0.6976,
+      "step": 399000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.630188331952846e-06,
+      "loss": 0.6659,
+      "step": 399500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.613455150891461e-06,
+      "loss": 0.6876,
+      "step": 400000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.596721969830076e-06,
+      "loss": 0.6982,
+      "step": 400500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.57998878876869e-06,
+      "loss": 0.7053,
+      "step": 401000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.563255607707304e-06,
+      "loss": 0.6955,
+      "step": 401500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.546522426645918e-06,
+      "loss": 0.7038,
+      "step": 402000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.529789245584533e-06,
+      "loss": 0.7101,
+      "step": 402500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.513056064523146e-06,
+      "loss": 0.68,
+      "step": 403000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.496322883461761e-06,
+      "loss": 0.6501,
+      "step": 403500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.479589702400375e-06,
+      "loss": 0.6824,
+      "step": 404000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.462856521338989e-06,
+      "loss": 0.7073,
+      "step": 404500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.4461233402776036e-06,
+      "loss": 0.7165,
+      "step": 405000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.429390159216218e-06,
+      "loss": 0.6787,
+      "step": 405500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.412656978154833e-06,
+      "loss": 0.7167,
+      "step": 406000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.395923797093447e-06,
+      "loss": 0.7228,
+      "step": 406500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.3791906160320616e-06,
+      "loss": 0.6817,
+      "step": 407000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.362457434970676e-06,
+      "loss": 0.6727,
+      "step": 407500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.34572425390929e-06,
+      "loss": 0.6911,
+      "step": 408000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.328991072847904e-06,
+      "loss": 0.6921,
+      "step": 408500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.312257891786519e-06,
+      "loss": 0.6894,
+      "step": 409000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.295524710725132e-06,
+      "loss": 0.7137,
+      "step": 409500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.278791529663747e-06,
+      "loss": 0.6994,
+      "step": 410000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.262058348602361e-06,
+      "loss": 0.7144,
+      "step": 410500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.245325167540975e-06,
+      "loss": 0.6939,
+      "step": 411000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.22859198647959e-06,
+      "loss": 0.6888,
+      "step": 411500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.211858805418205e-06,
+      "loss": 0.7098,
+      "step": 412000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.195125624356819e-06,
+      "loss": 0.72,
+      "step": 412500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.178392443295433e-06,
+      "loss": 0.7148,
+      "step": 413000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.1616592622340475e-06,
+      "loss": 0.6846,
+      "step": 413500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.144926081172662e-06,
+      "loss": 0.6869,
+      "step": 414000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.128192900111276e-06,
+      "loss": 0.7171,
+      "step": 414500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.11145971904989e-06,
+      "loss": 0.661,
+      "step": 415000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.094726537988505e-06,
+      "loss": 0.7095,
+      "step": 415500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.077993356927119e-06,
+      "loss": 0.6855,
+      "step": 416000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.061260175865733e-06,
+      "loss": 0.6964,
+      "step": 416500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.044526994804347e-06,
+      "loss": 0.6603,
+      "step": 417000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.027793813742963e-06,
+      "loss": 0.6788,
+      "step": 417500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.011060632681577e-06,
+      "loss": 0.6819,
+      "step": 418000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.994327451620191e-06,
+      "loss": 0.6967,
+      "step": 418500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.977594270558805e-06,
+      "loss": 0.7159,
+      "step": 419000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.96086108949742e-06,
+      "loss": 0.6882,
+      "step": 419500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9441279084360335e-06,
+      "loss": 0.6865,
+      "step": 420000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.927394727374648e-06,
+      "loss": 0.6902,
+      "step": 420500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9106615463132625e-06,
+      "loss": 0.6625,
+      "step": 421000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.893928365251876e-06,
+      "loss": 0.6674,
+      "step": 421500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.877195184190491e-06,
+      "loss": 0.6799,
+      "step": 422000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.860462003129105e-06,
+      "loss": 0.6669,
+      "step": 422500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.8437288220677205e-06,
+      "loss": 0.6894,
+      "step": 423000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.826995641006334e-06,
+      "loss": 0.6663,
+      "step": 423500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.810262459944949e-06,
+      "loss": 0.6763,
+      "step": 424000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.793529278883563e-06,
+      "loss": 0.6665,
+      "step": 424500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.776796097822177e-06,
+      "loss": 0.7065,
+      "step": 425000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.760062916760791e-06,
+      "loss": 0.6894,
+      "step": 425500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.743329735699406e-06,
+      "loss": 0.6467,
+      "step": 426000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.7265965546380195e-06,
+      "loss": 0.6896,
+      "step": 426500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.709863373576634e-06,
+      "loss": 0.7149,
+      "step": 427000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.6931301925152485e-06,
+      "loss": 0.6641,
+      "step": 427500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.676397011453862e-06,
+      "loss": 0.678,
+      "step": 428000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.659663830392477e-06,
+      "loss": 0.678,
+      "step": 428500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.642930649331092e-06,
+      "loss": 0.7005,
+      "step": 429000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.6261974682697065e-06,
+      "loss": 0.685,
+      "step": 429500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.60946428720832e-06,
+      "loss": 0.6912,
+      "step": 430000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.592731106146935e-06,
+      "loss": 0.6828,
+      "step": 430500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.575997925085549e-06,
+      "loss": 0.7087,
+      "step": 431000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.559264744024163e-06,
+      "loss": 0.6874,
+      "step": 431500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.542531562962777e-06,
+      "loss": 0.6931,
+      "step": 432000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.525798381901392e-06,
+      "loss": 0.6978,
+      "step": 432500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.509065200840006e-06,
+      "loss": 0.6559,
+      "step": 433000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.49233201977862e-06,
+      "loss": 0.6656,
+      "step": 433500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.475598838717234e-06,
+      "loss": 0.688,
+      "step": 434000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.45886565765585e-06,
+      "loss": 0.695,
+      "step": 434500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.442132476594464e-06,
+      "loss": 0.7068,
+      "step": 435000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.425399295533078e-06,
+      "loss": 0.695,
+      "step": 435500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.408666114471692e-06,
+      "loss": 0.703,
+      "step": 436000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.391932933410307e-06,
+      "loss": 0.6765,
+      "step": 436500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.3751997523489206e-06,
+      "loss": 0.67,
+      "step": 437000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.358466571287535e-06,
+      "loss": 0.6852,
+      "step": 437500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.3417333902261496e-06,
+      "loss": 0.7137,
+      "step": 438000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.325000209164763e-06,
+      "loss": 0.695,
+      "step": 438500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.308267028103378e-06,
+      "loss": 0.6873,
+      "step": 439000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.291533847041992e-06,
+      "loss": 0.6867,
+      "step": 439500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.274800665980606e-06,
+      "loss": 0.6912,
+      "step": 440000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.258067484919221e-06,
+      "loss": 0.6881,
+      "step": 440500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.241334303857836e-06,
+      "loss": 0.7207,
+      "step": 441000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.22460112279645e-06,
+      "loss": 0.6715,
+      "step": 441500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.207867941735064e-06,
+      "loss": 0.6639,
+      "step": 442000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.191134760673678e-06,
+      "loss": 0.6935,
+      "step": 442500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.174401579612293e-06,
+      "loss": 0.6972,
+      "step": 443000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.1576683985509065e-06,
+      "loss": 0.6989,
+      "step": 443500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.140935217489521e-06,
+      "loss": 0.6827,
+      "step": 444000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.1242020364281355e-06,
+      "loss": 0.6918,
+      "step": 444500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.107468855366749e-06,
+      "loss": 0.6746,
+      "step": 445000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.090735674305364e-06,
+      "loss": 0.6682,
+      "step": 445500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.074002493243979e-06,
+      "loss": 0.6935,
+      "step": 446000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0572693121825935e-06,
+      "loss": 0.695,
+      "step": 446500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.040536131121207e-06,
+      "loss": 0.6691,
+      "step": 447000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.023802950059822e-06,
+      "loss": 0.687,
+      "step": 447500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.007069768998436e-06,
+      "loss": 0.6904,
+      "step": 448000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.99033658793705e-06,
+      "loss": 0.6901,
+      "step": 448500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.973603406875664e-06,
+      "loss": 0.7144,
+      "step": 449000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.956870225814279e-06,
+      "loss": 0.6719,
+      "step": 449500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9401370447528925e-06,
+      "loss": 0.6511,
+      "step": 450000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.923403863691508e-06,
+      "loss": 0.7036,
+      "step": 450500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.906670682630122e-06,
+      "loss": 0.6645,
+      "step": 451000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.889937501568736e-06,
+      "loss": 0.6717,
+      "step": 451500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8732043205073505e-06,
+      "loss": 0.6898,
+      "step": 452000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.856471139445965e-06,
+      "loss": 0.6954,
+      "step": 452500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.839737958384579e-06,
+      "loss": 0.7217,
+      "step": 453000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.823004777323194e-06,
+      "loss": 0.6847,
+      "step": 453500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806271596261808e-06,
+      "loss": 0.6727,
+      "step": 454000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.789538415200422e-06,
+      "loss": 0.6785,
+      "step": 454500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.772805234139037e-06,
+      "loss": 0.6789,
+      "step": 455000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.75607205307765e-06,
+      "loss": 0.7096,
+      "step": 455500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.739338872016266e-06,
+      "loss": 0.6923,
+      "step": 456000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.722605690954879e-06,
+      "loss": 0.7319,
+      "step": 456500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.705872509893494e-06,
+      "loss": 0.6843,
+      "step": 457000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.689139328832108e-06,
+      "loss": 0.6602,
+      "step": 457500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.672406147770722e-06,
+      "loss": 0.6612,
+      "step": 458000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6556729667093365e-06,
+      "loss": 0.7042,
+      "step": 458500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.638939785647951e-06,
+      "loss": 0.6881,
+      "step": 459000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6222066045865655e-06,
+      "loss": 0.6487,
+      "step": 459500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.60547342352518e-06,
+      "loss": 0.6894,
+      "step": 460000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.588740242463794e-06,
+      "loss": 0.7034,
+      "step": 460500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.572007061402408e-06,
+      "loss": 0.6958,
+      "step": 461000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.555273880341023e-06,
+      "loss": 0.7096,
+      "step": 461500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.538540699279637e-06,
+      "loss": 0.693,
+      "step": 462000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.521807518218252e-06,
+      "loss": 0.7021,
+      "step": 462500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.505074337156865e-06,
+      "loss": 0.6756,
+      "step": 463000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.48834115609548e-06,
+      "loss": 0.7038,
+      "step": 463500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.471607975034094e-06,
+      "loss": 0.6838,
+      "step": 464000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.454874793972708e-06,
+      "loss": 0.666,
+      "step": 464500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.438141612911323e-06,
+      "loss": 0.6946,
+      "step": 465000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.421408431849937e-06,
+      "loss": 0.6724,
+      "step": 465500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.404675250788551e-06,
+      "loss": 0.6713,
+      "step": 466000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.387942069727166e-06,
+      "loss": 0.6862,
+      "step": 466500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3712088886657796e-06,
+      "loss": 0.6719,
+      "step": 467000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.354475707604395e-06,
+      "loss": 0.6763,
+      "step": 467500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3377425265430086e-06,
+      "loss": 0.6924,
+      "step": 468000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.321009345481623e-06,
+      "loss": 0.6841,
+      "step": 468500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3042761644202376e-06,
+      "loss": 0.6684,
+      "step": 469000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.287542983358852e-06,
+      "loss": 0.6851,
+      "step": 469500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.270809802297466e-06,
+      "loss": 0.6955,
+      "step": 470000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.254076621236081e-06,
+      "loss": 0.6943,
+      "step": 470500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.237343440174695e-06,
+      "loss": 0.695,
+      "step": 471000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.220610259113309e-06,
+      "loss": 0.6993,
+      "step": 471500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.203877078051924e-06,
+      "loss": 0.6954,
+      "step": 472000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.187143896990537e-06,
+      "loss": 0.6678,
+      "step": 472500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.170410715929152e-06,
+      "loss": 0.6946,
+      "step": 473000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.153677534867766e-06,
+      "loss": 0.6908,
+      "step": 473500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.136944353806381e-06,
+      "loss": 0.6732,
+      "step": 474000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.120211172744995e-06,
+      "loss": 0.6732,
+      "step": 474500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.103477991683609e-06,
+      "loss": 0.6523,
+      "step": 475000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0867448106222235e-06,
+      "loss": 0.6633,
+      "step": 475500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.070011629560838e-06,
+      "loss": 0.7044,
+      "step": 476000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0532784484994525e-06,
+      "loss": 0.6994,
+      "step": 476500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.036545267438067e-06,
+      "loss": 0.6906,
+      "step": 477000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.019812086376681e-06,
+      "loss": 0.6985,
+      "step": 477500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.003078905315295e-06,
+      "loss": 0.6817,
+      "step": 478000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.98634572425391e-06,
+      "loss": 0.6997,
+      "step": 478500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.969612543192524e-06,
+      "loss": 0.6872,
+      "step": 479000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.952879362131139e-06,
+      "loss": 0.6919,
+      "step": 479500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.936146181069752e-06,
+      "loss": 0.6785,
+      "step": 480000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.919413000008367e-06,
+      "loss": 0.671,
+      "step": 480500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.902679818946981e-06,
+      "loss": 0.6932,
+      "step": 481000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.885946637885595e-06,
+      "loss": 0.6747,
+      "step": 481500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.86921345682421e-06,
+      "loss": 0.6812,
+      "step": 482000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.852480275762824e-06,
+      "loss": 0.6634,
+      "step": 482500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8357470947014385e-06,
+      "loss": 0.6858,
+      "step": 483000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.819013913640053e-06,
+      "loss": 0.6444,
+      "step": 483500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.802280732578667e-06,
+      "loss": 0.7009,
+      "step": 484000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.785547551517281e-06,
+      "loss": 0.6686,
+      "step": 484500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.768814370455896e-06,
+      "loss": 0.6901,
+      "step": 485000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.75208118939451e-06,
+      "loss": 0.6858,
+      "step": 485500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7353480083331246e-06,
+      "loss": 0.6713,
+      "step": 486000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7186148272717387e-06,
+      "loss": 0.6853,
+      "step": 486500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.701881646210353e-06,
+      "loss": 0.6969,
+      "step": 487000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6851484651489673e-06,
+      "loss": 0.6564,
+      "step": 487500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.668415284087582e-06,
+      "loss": 0.659,
+      "step": 488000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6516821030261963e-06,
+      "loss": 0.6865,
+      "step": 488500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6349489219648104e-06,
+      "loss": 0.6807,
+      "step": 489000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6182157409034244e-06,
+      "loss": 0.6564,
+      "step": 489500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.601482559842039e-06,
+      "loss": 0.6816,
+      "step": 490000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5847493787806534e-06,
+      "loss": 0.7028,
+      "step": 490500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.568016197719268e-06,
+      "loss": 0.665,
+      "step": 491000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.551283016657882e-06,
+      "loss": 0.6886,
+      "step": 491500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5345498355964965e-06,
+      "loss": 0.6722,
+      "step": 492000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5178166545351106e-06,
+      "loss": 0.678,
+      "step": 492500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5010834734737247e-06,
+      "loss": 0.6729,
+      "step": 493000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4843502924123396e-06,
+      "loss": 0.6765,
+      "step": 493500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4676171113509537e-06,
+      "loss": 0.6857,
+      "step": 494000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.450883930289568e-06,
+      "loss": 0.6695,
+      "step": 494500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4341507492281823e-06,
+      "loss": 0.6515,
+      "step": 495000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4174175681667963e-06,
+      "loss": 0.6873,
+      "step": 495500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.400684387105411e-06,
+      "loss": 0.6656,
+      "step": 496000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3839512060440253e-06,
+      "loss": 0.6858,
+      "step": 496500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.36721802498264e-06,
+      "loss": 0.6596,
+      "step": 497000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.350484843921254e-06,
+      "loss": 0.7077,
+      "step": 497500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.333751662859868e-06,
+      "loss": 0.6677,
+      "step": 498000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3170184817984825e-06,
+      "loss": 0.6562,
+      "step": 498500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3002853007370966e-06,
+      "loss": 0.6433,
+      "step": 499000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2835521196757115e-06,
+      "loss": 0.6926,
+      "step": 499500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2668189386143256e-06,
+      "loss": 0.6654,
+      "step": 500000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.25008575755294e-06,
+      "loss": 0.66,
+      "step": 500500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.233352576491554e-06,
+      "loss": 0.6735,
+      "step": 501000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2166193954301682e-06,
+      "loss": 0.6818,
+      "step": 501500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.199886214368783e-06,
+      "loss": 0.6919,
+      "step": 502000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1831530333073972e-06,
+      "loss": 0.6881,
+      "step": 502500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1664198522460117e-06,
+      "loss": 0.6652,
+      "step": 503000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.149686671184626e-06,
+      "loss": 0.6749,
+      "step": 503500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.13295349012324e-06,
+      "loss": 0.6808,
+      "step": 504000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1162203090618544e-06,
+      "loss": 0.6882,
+      "step": 504500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.099487128000469e-06,
+      "loss": 0.6713,
+      "step": 505000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0827539469390834e-06,
+      "loss": 0.6607,
+      "step": 505500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0660207658776974e-06,
+      "loss": 0.6789,
+      "step": 506000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0492875848163115e-06,
+      "loss": 0.6786,
+      "step": 506500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.032554403754926e-06,
+      "loss": 0.6703,
+      "step": 507000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.01582122269354e-06,
+      "loss": 0.6573,
+      "step": 507500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.999088041632155e-06,
+      "loss": 0.6848,
+      "step": 508000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.982354860570769e-06,
+      "loss": 0.7117,
+      "step": 508500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9656216795093836e-06,
+      "loss": 0.6853,
+      "step": 509000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9488884984479977e-06,
+      "loss": 0.6834,
+      "step": 509500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9321553173866117e-06,
+      "loss": 0.6913,
+      "step": 510000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9154221363252262e-06,
+      "loss": 0.6746,
+      "step": 510500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8986889552638407e-06,
+      "loss": 0.6813,
+      "step": 511000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8819557742024553e-06,
+      "loss": 0.6621,
+      "step": 511500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8652225931410693e-06,
+      "loss": 0.6781,
+      "step": 512000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8484894120796834e-06,
+      "loss": 0.657,
+      "step": 512500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.831756231018298e-06,
+      "loss": 0.6414,
+      "step": 513000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8150230499569124e-06,
+      "loss": 0.6879,
+      "step": 513500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.798289868895527e-06,
+      "loss": 0.6901,
+      "step": 514000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.781556687834141e-06,
+      "loss": 0.6908,
+      "step": 514500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.764823506772755e-06,
+      "loss": 0.6905,
+      "step": 515000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7480903257113696e-06,
+      "loss": 0.6663,
+      "step": 515500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7313571446499836e-06,
+      "loss": 0.654,
+      "step": 516000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7146239635885986e-06,
+      "loss": 0.6968,
+      "step": 516500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6978907825272126e-06,
+      "loss": 0.6675,
+      "step": 517000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.681157601465827e-06,
+      "loss": 0.6779,
+      "step": 517500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.664424420404441e-06,
+      "loss": 0.6933,
+      "step": 518000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6476912393430553e-06,
+      "loss": 0.6667,
+      "step": 518500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6309580582816698e-06,
+      "loss": 0.6941,
+      "step": 519000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6142248772202843e-06,
+      "loss": 0.6485,
+      "step": 519500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5974916961588988e-06,
+      "loss": 0.6815,
+      "step": 520000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.580758515097513e-06,
+      "loss": 0.6738,
+      "step": 520500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.564025334036127e-06,
+      "loss": 0.6712,
+      "step": 521000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5472921529747414e-06,
+      "loss": 0.7071,
+      "step": 521500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5305589719133555e-06,
+      "loss": 0.6649,
+      "step": 522000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5138257908519704e-06,
+      "loss": 0.6604,
+      "step": 522500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4970926097905845e-06,
+      "loss": 0.6927,
+      "step": 523000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4803594287291986e-06,
+      "loss": 0.6728,
+      "step": 523500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.463626247667813e-06,
+      "loss": 0.6821,
+      "step": 524000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4468930666064276e-06,
+      "loss": 0.6638,
+      "step": 524500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4301598855450417e-06,
+      "loss": 0.6783,
+      "step": 525000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4134267044836557e-06,
+      "loss": 0.6808,
+      "step": 525500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3966935234222707e-06,
+      "loss": 0.667,
+      "step": 526000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3799603423608847e-06,
+      "loss": 0.6607,
+      "step": 526500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.363227161299499e-06,
+      "loss": 0.6571,
+      "step": 527000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3464939802381133e-06,
+      "loss": 0.6699,
+      "step": 527500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.329760799176728e-06,
+      "loss": 0.6613,
+      "step": 528000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.313027618115342e-06,
+      "loss": 0.7355,
+      "step": 528500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2962944370539564e-06,
+      "loss": 0.672,
+      "step": 529000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2795612559925705e-06,
+      "loss": 0.6961,
+      "step": 529500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.262828074931185e-06,
+      "loss": 0.6712,
+      "step": 530000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2460948938697995e-06,
+      "loss": 0.6757,
+      "step": 530500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2293617128084136e-06,
+      "loss": 0.6664,
+      "step": 531000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.212628531747028e-06,
+      "loss": 0.6705,
+      "step": 531500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.195895350685642e-06,
+      "loss": 0.6799,
+      "step": 532000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1791621696242566e-06,
+      "loss": 0.6822,
+      "step": 532500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.162428988562871e-06,
+      "loss": 0.6833,
+      "step": 533000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.145695807501485e-06,
+      "loss": 0.6869,
+      "step": 533500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1289626264400993e-06,
+      "loss": 0.6556,
+      "step": 534000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1122294453787138e-06,
+      "loss": 0.6761,
+      "step": 534500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0954962643173283e-06,
+      "loss": 0.6857,
+      "step": 535000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0787630832559424e-06,
+      "loss": 0.671,
+      "step": 535500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.062029902194557e-06,
+      "loss": 0.6775,
+      "step": 536000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0452967211331714e-06,
+      "loss": 0.6568,
+      "step": 536500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0285635400717854e-06,
+      "loss": 0.6763,
+      "step": 537000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0118303590104e-06,
+      "loss": 0.6832,
+      "step": 537500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.995097177949014e-06,
+      "loss": 0.671,
+      "step": 538000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9783639968876285e-06,
+      "loss": 0.6607,
+      "step": 538500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.961630815826243e-06,
+      "loss": 0.6783,
+      "step": 539000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.944897634764857e-06,
+      "loss": 0.6529,
+      "step": 539500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.928164453703471e-06,
+      "loss": 0.6999,
+      "step": 540000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9114312726420857e-06,
+      "loss": 0.6939,
+      "step": 540500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8946980915807e-06,
+      "loss": 0.6845,
+      "step": 541000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8779649105193147e-06,
+      "loss": 0.6757,
+      "step": 541500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8612317294579287e-06,
+      "loss": 0.6742,
+      "step": 542000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.844498548396543e-06,
+      "loss": 0.6694,
+      "step": 542500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8277653673351575e-06,
+      "loss": 0.6902,
+      "step": 543000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8110321862737718e-06,
+      "loss": 0.7054,
+      "step": 543500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.794299005212386e-06,
+      "loss": 0.6628,
+      "step": 544000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7775658241510004e-06,
+      "loss": 0.675,
+      "step": 544500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7608326430896147e-06,
+      "loss": 0.676,
+      "step": 545000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.744099462028229e-06,
+      "loss": 0.6773,
+      "step": 545500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7273662809668435e-06,
+      "loss": 0.6773,
+      "step": 546000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7106330999054578e-06,
+      "loss": 0.6828,
+      "step": 546500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6938999188440718e-06,
+      "loss": 0.7134,
+      "step": 547000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6771667377826863e-06,
+      "loss": 0.6625,
+      "step": 547500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6604335567213006e-06,
+      "loss": 0.6744,
+      "step": 548000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.643700375659915e-06,
+      "loss": 0.6694,
+      "step": 548500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6269671945985294e-06,
+      "loss": 0.6706,
+      "step": 549000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6102340135371435e-06,
+      "loss": 0.6701,
+      "step": 549500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5935008324757578e-06,
+      "loss": 0.6655,
+      "step": 550000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5767676514143723e-06,
+      "loss": 0.6588,
+      "step": 550500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5600344703529866e-06,
+      "loss": 0.6739,
+      "step": 551000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5433012892916009e-06,
+      "loss": 0.6462,
+      "step": 551500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5265681082302154e-06,
+      "loss": 0.6706,
+      "step": 552000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5098349271688294e-06,
+      "loss": 0.6935,
+      "step": 552500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4931017461074437e-06,
+      "loss": 0.667,
+      "step": 553000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4763685650460582e-06,
+      "loss": 0.675,
+      "step": 553500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4596353839846725e-06,
+      "loss": 0.6553,
+      "step": 554000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.442902202923287e-06,
+      "loss": 0.7021,
+      "step": 554500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4261690218619013e-06,
+      "loss": 0.6511,
+      "step": 555000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4094358408005154e-06,
+      "loss": 0.6849,
+      "step": 555500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3927026597391299e-06,
+      "loss": 0.6599,
+      "step": 556000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3759694786777442e-06,
+      "loss": 0.6564,
+      "step": 556500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3592362976163585e-06,
+      "loss": 0.6941,
+      "step": 557000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.342503116554973e-06,
+      "loss": 0.652,
+      "step": 557500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.325769935493587e-06,
+      "loss": 0.687,
+      "step": 558000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3090367544322013e-06,
+      "loss": 0.6653,
+      "step": 558500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2923035733708158e-06,
+      "loss": 0.6701,
+      "step": 559000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2755703923094301e-06,
+      "loss": 0.6589,
+      "step": 559500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2588372112480444e-06,
+      "loss": 0.665,
+      "step": 560000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2421040301866587e-06,
+      "loss": 0.6581,
+      "step": 560500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.225370849125273e-06,
+      "loss": 0.6885,
+      "step": 561000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2086376680638875e-06,
+      "loss": 0.6603,
+      "step": 561500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1919044870025018e-06,
+      "loss": 0.6647,
+      "step": 562000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.175171305941116e-06,
+      "loss": 0.6693,
+      "step": 562500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1584381248797303e-06,
+      "loss": 0.6577,
+      "step": 563000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1417049438183448e-06,
+      "loss": 0.6608,
+      "step": 563500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.124971762756959e-06,
+      "loss": 0.6926,
+      "step": 564000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1082385816955734e-06,
+      "loss": 0.6693,
+      "step": 564500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0915054006341877e-06,
+      "loss": 0.6567,
+      "step": 565000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.074772219572802e-06,
+      "loss": 0.6646,
+      "step": 565500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0580390385114163e-06,
+      "loss": 0.6811,
+      "step": 566000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0413058574500306e-06,
+      "loss": 0.6873,
+      "step": 566500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0245726763886449e-06,
+      "loss": 0.6673,
+      "step": 567000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0078394953272591e-06,
+      "loss": 0.6689,
+      "step": 567500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.911063142658736e-07,
+      "loss": 0.6832,
+      "step": 568000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.74373133204488e-07,
+      "loss": 0.6649,
+      "step": 568500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.576399521431022e-07,
+      "loss": 0.6641,
+      "step": 569000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.409067710817166e-07,
+      "loss": 0.6714,
+      "step": 569500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.241735900203309e-07,
+      "loss": 0.664,
+      "step": 570000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.074404089589452e-07,
+      "loss": 0.6681,
+      "step": 570500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.907072278975596e-07,
+      "loss": 0.6392,
+      "step": 571000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.739740468361739e-07,
+      "loss": 0.6888,
+      "step": 571500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.572408657747882e-07,
+      "loss": 0.6505,
+      "step": 572000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.405076847134025e-07,
+      "loss": 0.6462,
+      "step": 572500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.237745036520168e-07,
+      "loss": 0.6713,
+      "step": 573000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.070413225906312e-07,
+      "loss": 0.6539,
+      "step": 573500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.903081415292454e-07,
+      "loss": 0.6516,
+      "step": 574000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.735749604678598e-07,
+      "loss": 0.6652,
+      "step": 574500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.568417794064742e-07,
+      "loss": 0.6537,
+      "step": 575000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.401085983450884e-07,
+      "loss": 0.6674,
+      "step": 575500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.233754172837028e-07,
+      "loss": 0.6718,
+      "step": 576000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.066422362223172e-07,
+      "loss": 0.6772,
+      "step": 576500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.899090551609314e-07,
+      "loss": 0.6954,
+      "step": 577000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.731758740995458e-07,
+      "loss": 0.6841,
+      "step": 577500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.564426930381602e-07,
+      "loss": 0.6886,
+      "step": 578000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.397095119767743e-07,
+      "loss": 0.6778,
+      "step": 578500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.229763309153887e-07,
+      "loss": 0.6625,
+      "step": 579000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.06243149854003e-07,
+      "loss": 0.6562,
+      "step": 579500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.895099687926174e-07,
+      "loss": 0.6649,
+      "step": 580000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.727767877312317e-07,
+      "loss": 0.6976,
+      "step": 580500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.56043606669846e-07,
+      "loss": 0.6579,
+      "step": 581000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.393104256084603e-07,
+      "loss": 0.6331,
+      "step": 581500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.225772445470747e-07,
+      "loss": 0.6964,
+      "step": 582000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.05844063485689e-07,
+      "loss": 0.6688,
+      "step": 582500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.891108824243032e-07,
+      "loss": 0.6594,
+      "step": 583000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7237770136291764e-07,
+      "loss": 0.6663,
+      "step": 583500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5564452030153193e-07,
+      "loss": 0.6763,
+      "step": 584000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.389113392401463e-07,
+      "loss": 0.654,
+      "step": 584500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.221781581787606e-07,
+      "loss": 0.6605,
+      "step": 585000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.054449771173749e-07,
+      "loss": 0.6639,
+      "step": 585500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.887117960559893e-07,
+      "loss": 0.6615,
+      "step": 586000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.719786149946036e-07,
+      "loss": 0.6779,
+      "step": 586500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.552454339332179e-07,
+      "loss": 0.6681,
+      "step": 587000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.385122528718322e-07,
+      "loss": 0.6806,
+      "step": 587500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.217790718104465e-07,
+      "loss": 0.6689,
+      "step": 588000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0504589074906084e-07,
+      "loss": 0.6651,
+      "step": 588500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.883127096876752e-07,
+      "loss": 0.6824,
+      "step": 589000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.715795286262895e-07,
+      "loss": 0.6882,
+      "step": 589500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5484634756490387e-07,
+      "loss": 0.6956,
+      "step": 590000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3811316650351818e-07,
+      "loss": 0.678,
+      "step": 590500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2137998544213247e-07,
+      "loss": 0.6782,
+      "step": 591000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.046468043807468e-07,
+      "loss": 0.6526,
+      "step": 591500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8791362331936115e-07,
+      "loss": 0.6804,
+      "step": 592000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7118044225797547e-07,
+      "loss": 0.6891,
+      "step": 592500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5444726119658978e-07,
+      "loss": 0.6855,
+      "step": 593000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3771408013520412e-07,
+      "loss": 0.6647,
+      "step": 593500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2098089907381844e-07,
+      "loss": 0.6531,
+      "step": 594000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0424771801243277e-07,
+      "loss": 0.6683,
+      "step": 594500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.751453695104708e-08,
+      "loss": 0.6539,
+      "step": 595000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.078135588966141e-08,
+      "loss": 0.687,
+      "step": 595500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.404817482827574e-08,
+      "loss": 0.6707,
+      "step": 596000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.731499376689006e-08,
+      "loss": 0.675,
+      "step": 596500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.058181270550438e-08,
+      "loss": 0.6697,
+      "step": 597000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.848631644118706e-09,
+      "loss": 0.6906,
+      "step": 597500
+    }
+  ],
+  "max_steps": 597615,
+  "num_train_epochs": 1,
+  "total_flos": 4.809254464703693e+16,
+  "trial_name": null,
+  "trial_params": null
+}